2-DPR: A Novel, High Performance Cache Replacement Algorithm

Abstract

Caching is a fundamental technique commonly employed to hide the latency gap between memory and the CPU by exploiting locality in memory accesses. Different cache replacement algorithms have dramatically different effects on the system performance by deciding which blocks to evict from cache memory in case of a cache miss occurs. The aim of these replacement techniques is trying to get closer to the optimal case by achieving best usage of the total size of the cache, minimizing the miss ratio as much as possible and accomplishing the highest system performance can be reached. In this paper, a simple and elegant new algorithm is proposed, namely, Two-Dimensional Pyramid Replacement, (2-DPR), that combines the advantages of (LRU) and (LFU), and eliminates their disadvantages.